def DENCLUE(D, h, kexi, eps):
A = list() #吸引子集合
R = dict() #被吸引子吸引的点的集合
for i in range(D.shape[0]):
x = D[i]
Xxing = FINDATTRACTOR(x, D, h, eps)
# Xxing=getround(Xxing)
Xxingli = list(Xxing)
# if fx(Xxing,D,h)>=kexi:
if fx(Xxing, D, h) >= kexi:
A.append(Xxingli)
if str(Xxingli) not in R.keys():
a = []
a.append(x)
R[str(Xxingli)] = a
else:
a = R[str(Xxingli)]
a.append(x)
R[str(Xxingli)] = a
#eps1 距离的阈值
#min_sample要成为核心对象所需要的ϵ-邻域的样本数阈值
print("A", A)
# A=list(set(A))
eps1 = 0.1
min_sample = 2
a = DBSCAN(eps=eps1, min_samples=min_sample)
HC = a.fit_predict(np.mat(A)) #dbscan聚类结果
clusterset = set()
clusterdict = dict()
for i in range(len(HC)):
if (HC[i] != -1):
clusterset.add(HC[i])
if str(HC[i]) not in clusterdict.keys():
a = []
a.append(A[i])
clusterdict[str(HC[i])] = a
else:
a = clusterdict[str(HC[i])]
a.append(A[i])
print("number of clusters", len(clusterdict.keys()))
for key in R.keys():
print("density attracotr", key)
print("atracted point", R[key])
for i in clusterdict.keys():
ls = clusterdict[i]
for x in clusterdict[i]:
if str(x) in R.keys():
for z in R[str(x)]:
ls.append(z)
clusterdict[i] = ls
for i in clusterdict.keys():
print('clusterres', i, " size", len(clusterdict[i]))
print('set of point in the cluster', clusterdict[i])
showpicture(clusterdict)
def denclue(points, bandwidth, hyper, convergence):
A = list() # 定义一个列表存吸引子
r = dict() # 定义一个字典存吸引子对应的坐标值
for i in range(points.shape[0]): # 对点集中的每个点判断
x = findAttractor(bandwidth, convergence, points[i],
points) # 求出每个点的吸引子
# 用密度估计判断这个吸引子，吸引的点在区域内的密度大小。如果太小不足以构成一类就舍弃
if multi(x, points, bandwidth) > hyper:
A.append(list(x)) # 像列表A中插入符合条件的吸引子
strxing = str(x) # 将吸引子的坐标变成字符串
if strxing not in r.keys(): # 如果字典中没有这个吸引子的坐标值
a = [] # 定义一个数组
a.append(points[i]) # 将对应的点全部加入
r[strxing] = a # 将吸引子的值与这些点对应
# 有这个值，那就将点直接加入，一个吸引子可以对应很多点
else:
a = r[strxing]
a.append(points[i])
r[strxing] = a
E = 0.1
min = 2
b = DBSCAN(eps=E, min_samples=min) # 用DBCSACN进行聚类，密度可达，最大化
c = b.fit_predict(np.mat(list(A))) # 训练之后的结果
b_set = set()
b_dict = dict()
for i in range(len(c)): # 将训练完之后的类加入集合
if (c[i] != -1):
b_set.add(c[i])
if str(c[i]) not in b_dict.keys():
b = []
b.append(A[i])
b_dict[str(c[i])] = b
else:
b = b_dict[str(c[i])]
b.append(A[i])
print("clusters number", len(b_dict.keys())) # 输出每个类及其中的点
for i in b_dict.keys(): # 对每一类进行判断，如果吸引子属于该类则加入
xs = b_dict[i]
for x in b_dict[i]:
if str(x) in r.keys():
for i in r[str(x)]:
xs.append(i)
b_dict[i] = xs
for i in b_dict.keys():
print("群集", i)
print("群集", i, "中的点", b_dict[i])
